PURPOSE: To automatically detect the defective cutting at a welded steel tube cut section on the on-line basis by fine line processing a cross section-shaped image, extracting the cut section via intensity, and comparing an approximate curve with the cut cross sectional shape. ;CONSTITUTION: A steel pipe 5 is arranged so that the step shape of a cut face immediately after being cut by a cutting tool 6 can be recognized as an image. A slit light projector 1 is installed at the prescribed angle against the steel pipe 5, and a television camera 2 is installed at a position perpendicular to the steel pipe 5. The slit light 3 from the projector 1 is radiated to the outer cut face of the steel pipe 5, and the obtained outer cut cross sectional image is photographed by the television camera 2. The outer cut cross sectional image is inputted to an image processing device 7, the image is fine line processed, and a cut section is extracted by the cross sectional shape intensity distribution difference. The cut section cross sectional shape wave-form is secondarily approximated, the obtained approximate curve is compared with the cut section cross sectional shape wave-form, and the local irregular defect in the cut section is detected. When the defect is detected, it is reported by a warning device 8 and a marking device 9.;COPYRIGHT: (C)1994,JPO&Japio
展开▼